New AI Software May Key Successful Antiterrorism, Military and Diplomatic Actions


Newswise — Six years after the 911 attacks, Al Qaeda, and the Taliban have regrouped. The resurgence of these groups, the war in Iraq, and a surge in conflict around the world, all emphasize the critical need for computer technologies that can provide rapid information on terrorists or on the cultural and political climate on the ground in regions of critical interest. The University of Maryland Institute for Advanced Computer Studies (UMIACS) has developed computer analysis software for these applications.

A Perspectives piece by UMIACS Director V.S. Subrahmanian in the September 14 issue of the journal Science looks at these new computer models and databases that can help policymakers or military planners predict the behavior of political, economic, and social groups, predictions that can be critical for success. In his article, Subrahmanian notes that in 2001 U.S., commanders probably knew where bin Laden was, but were unable to prevent his escape, in part because inadequate cultural knowledge prevented troops on the ground from successfully negotiating with local tribesmen. According to Subrahmanian, such failures can be prevented if decision makers have access to all pertinent data and to accurate models of the behaviors of the relevant groups.

Subrahmanian and his colleagues at Maryland have developed a host of software tools to track information about foreign groups in sources ranging from news sources to blogs to online video libraries. This software can almost instantly search the entire internet for information/links for a terrorist suspect or other particular person, group, etc. of interest. Moreover, in conjunction with social scientists at the University of Maryland who have studied world trouble spots (http://www.newsdesk.umd.edu/sociss/release.cfm?ArticleID=1491), UMIACS computer scientists have developed methods to extract rules governing the behaviors of different groups in foreign countries. As an example, they have extracted about 14,000 rules about Hezbollah alone.

"Currently we are able to find the most probable sets of actions a group could take when there are up to 10 to the power of 27 (i.e. 10 with 27 zeros after it) possible sets of actions the group can take," says Subrahmanian. "Compare this with the number of atoms on earth which is about 10 to the power of 50."

See earlier AAAS news about this work at http://www.aaas.org/news/releases/2007/0625insurgents.shtml or http://www.sciencemag.org/cgi/content/full/316/5824/534?maxtoshow=&HITS=.

  • share-facebook-New AI Software May Key Successful Antiterrorism, Military and Diplomatic Actions
  • share-twitter-New AI Software May Key Successful Antiterrorism, Military and Diplomatic Actions

Comment/Share

step 2
Chat now!